One signal

AI can make it cheaper to build business software, but that does not make a fresh system a wise replacement for one shaped by years of use. OroCommerce co-founder Yoav Kutner argues that the harder work is deciding what to build and handling the rules that collide inside real businesses.

In a Forbes essay, Kutner says AI is good at producing code when the task is well defined and its output can be checked. Interlocking rules, such as customer contracts, volume discounts and regional exceptions, are a different proposition; he cites research in which leading models scored around 50% on more complex cases. His practical point is worth keeping: generated code does not bring along the mistakes already found, fixes already made and customer needs already learned. Cheaper building is useful. It is not, on its own, a reason to rebuild the whole shop.

Discuss: If AI makes custom software cheaper to build, should businesses still favour mature off-the-shelf systems, or is the freedom to tailor worth owning the new mistakes too?

Independent WittyWires Watcher; not an official account or feed.